Re: [ECOS] Problem while Performing SoftReset when exception occurs

[Jonathan Larmour <jifl at eCosCentric dot com>]

Gary D. Thomas wrote:
> On Fri, 2003-02-14 at 22:39, venkatadri potluri wrote:
>
> > i am using MPC860 based board. when ever exception occurs i want to reset
> > the board.i am generating CYGNUM_HAL_VECTOR_DTLB_MISS exception for
> > testing.
>
> Have you tried calling HAL_PLATFORM_RESET() in your handler?  This 
> should invoke a sequence which resets the platform and is much safer
> than just calling the reset handler.
If HAL_PLATFORM_RESET() isn't defined (or for the MPC8xx if you don't have 
a QUICC, since it just loops forever otherwise, which is somewhat bogus 
actually - it just shouldn't be defined instead) then branching to the 
start address _may_ be okay but there are some things I saw you 
(Venkatadri) missed:

> > The problem i am facing is when ever exception occurs control is going to
> > exception handler and then to _exception_reset, up to here it is OK. While
> > booting up ecos again control is hanging in cyg_hal_invoke_constuctors().
> > if we debug inside cyg_hal_invoke_constuctors(), control path is like
> > this
> >
> >
> >
> > Cyg_Exception_Control::Cyg_Exception_Control
> >                    |
> > 		    |
> >       cyg_hal_default_exception_vsr
> >                    |
> >                    |
> >        cyg_hal_exception_handler
> >                    |
> > 		    |
> > Cyg_Exception_Control::deliver_exception
> >
> >
> > and control is switching between cyg_hal_exception_handler and
> > deliver_exception.  i guess  exception was not cleared properly.
> >
> > so how to clear the  exception or solve this problem..
Look at the powerPC exception handler in vectors.S, specifically 
restore_state. You should also disable interrupts and disable the MMU. 
Otherwise the system may not be in the state it expects (depends on the 
specific platform setup code admittedly) and in particular you may get an 
interrupt when you're not in a state to receive them.

Finally, if this is a RAM startup app rather than something booted 
straight from ROM/Flash stuff like the data section won't have been reset. 
If that's what you want you'd either have to reload the data section or 
even stash away a copy of the initial state of the data section at boot time!

But yes, if at all possible use HAL_PLATFORM_RESET, or if that doesn't 
work, some other hardware watchdog.

Jifl
--
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine


--
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss